Exploring the possibilities of genetic improvement from traceability data

Livestock Science 125 (2009) 115–120

Contents lists available at ScienceDirect

Livestock Science j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / l i v s c i

Exploring the possibilities of genetic improvement from traceability data An example in the Pirenaica beef cattle J. Altarriba ⁎, G. Yagüe, C. Moreno, L. Varona Genética cuantitativa y Mejora animal, Facultad de Veterinaria, Universidad de Zaragoza, E-50013 Zaragoza, Spain

a r t i c l e

i n f o

Article history: Received 3 October 2008 Received in revised form 25 March 2009 Accepted 27 March 2009 Keywords: Databases Carcass weight Conformation Fat cover Colour Weaning Weight Heritability Genetic correlation

a b s t r a c t The aim of this study is to explore the potential use of the information generated by the Spanish traceability program (SIMOGAN) for animal breeding purposes in the Pirenaica beef cattle breed. The traits included in the study were: cold carcass weight (CW, n = 20,010), conformation (CON, n = 15,808), fat cover (FC, n = 13,739) and colour (COL, n = 3477) from the SIMOGAN database; and weaning weight (WW, n = 15,561) from the Breeders Association (CONASPI) database. Posterior marginal estimates of genetic parameters were obtained using Bayesian inference, implemented via a Gibbs sampling scheme. Posterior marginal means of heritabilities were 0.34, 0.28, 0.19, 0.23 and 0.38 for CW, CON, FC, COL and WW, respectively. Moreover, posterior marginal distributions of genetic correlations between CW-CON, CW-WW, CON-FC and FC-WW do not include the zero within the Highest Posterior Density (HPD) at 95%, and their posterior mean estimates were 0.30, 0.54, − 0.35 and 0.23, respectively. These results indicate that there is enough genetic variability for selection in CW, CON, FC, COL and WW. The availability of records is potentially abundant at a very low cost, thus they can be easily included in the selection criteria. Consequences of the current selection criteria (WW) and other possible alternatives are discussed. © 2009 Elsevier B.V. All rights reserved.

1. Introduction In recent years, new information systems have been developed to guarantee the traceability of animal products and veterinary epidemiological surveillance. In particular, the European Union has obliged to each Member State to create a computerized database to record the identity of all animals, all holdings on their territory and the movements of every individual from data generated in farms, slaughterhouses, etc. (Council Directive 97/12/EC, 1997). In Spain, according to this demand, the traceability SIMOGAN database (Sistema de Identificación y Movimiento de Ganado Bovino — National System of Identification and Registration the Movements of the Bovines) was set up (http://www.mapa.es/es/ganaderia/pags/simogan/simogan.htm#art2) to record the

⁎ Corresponding author. Facultad de Veterinaria, Universidad de Zaragoza, C. Miguel Servet, 177, E-50013 Zaragoza, Spain. Tel.: +34 976 761623; fax: +34 976 761612. E-mail address: [email protected] (J. Altarriba). 1871-1413/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.livsci.2009.03.013

movement of all animals with the aim of facilitating intracommunity trade and the epidemic pursuit of the animal populations (BOE, 1999). The Pirenaica beef cattle breed is an extensive population from northern Spain that consists of about 20,000 individuals (Sanchez et al., 2002). The phenotypic information recorded by the SIMOGAN database, at the slaughterhouse level, consists of carcass weight (CW), carcass conformation (CON), fat cover (FC) and meat colour (COL). From a different point of view, since 1988, a selection program has been applied by the Pirenaica Breeders Association (CONASPI) and based on a weaning weight (WW) index, calculated from weights recorded between 110 and 310 days of age (Altarriba et al., 1996; Varona et al., 1997). Weaning weight can be recorded on the candidates for selection, but potential response to selection should be generated only through the correlated response with CW. However, the SIMOGAN database provides direct information from the actual objective (CW), although the phenotypes for the selection candidates are not available.

116

J. Altarriba et al. / Livestock Science 125 (2009) 115–120

Furthermore, the information generated by the SIMOGAN database opens the possibility of including traits related with meat and carcass quality in the selection criteria. These traits are becoming more important in beef cattle production (Eriksson et al., 2003), but, until now, their practical implementation in selection schemes has been constrained by the limited availability of recorded data in a substantial subset of the population. The regular measurement of carcass and meat quality traits in a large population is extremely expensive, prohibitively so for animal breeding practices. Until now, most of the studies that characterize European beef cattle populations have been based on reduced sample sizes (Piedrafita et al., 2003; Renand et al., 2003; Albertí et al., 2005), and the information generated has been useful for genetic improvement with some special statistical designs (Altarriba et al., 2005, 2006). The objective of this study was to explore the possibilities of implementing new selection criteria for the Pirenaica breed, given the information provided by the SIMOGAN database, and to study the consequences in meat and carcass quality traits by the estimation of the (co)variance components and the expected direct and correlated response.

2. Materials and methods 2.1. Data The individuals included in this analysis were purebred Pirenaica yearling calves slaughtered between 1999 and 2007 in 12 slaughterhouses located in Basque Country and Navarre (Spain). The traits contributed by SIMOGAN database to this study were: cold carcass weight (CW), conformation (CON), in function of the development of essential parts of carcass profile according to the (S)EUROP scale (CEE no 2930/81, 1981), fat cover score (FC), quantifying the amount of fat on the outside of the carcass and in the thoracic cavity, and beef colour (COL) scored from 1 (light) to 7 (dark) according to the Beef Color Standard (BCS). The categorical scale of CON was converted to a numeric scale from 1.00 (P) to 6.00 (S) with 16 possible values separated by increments of 0.33. Similarly, FC and COL were converted into 17 possible scores between 1.00 and 5.00. Scores over 5 were not observed for COL. The data were completed with pedigree and weaning weight records at 210 days from 1989 to 2007 provided by the Pirenaica Breeders Association. Both CONASPI and SIMOGAN databases were merged according the European animal identification code by CONASPI. A summary of the descriptive statistics of the phenotypic traits is presented in Table 1.

Table 1 Mean (± SE) and coefficient of variation (CV) of the studied traits. Variable a

CW

CON

FC

COL

WW

N Mean CV (%)

20,010 297.7 ± 2.1 18.5

15,808 3.60 ± 0.03 12.9

13,739 2.16 ± 0.02 23.8

3477 2.26 ± 0.04 14.9

15,561 264.8 ± 2.1 22.2

a CW, cold carcass weight in kg; CON, conformation; FC, fat cover; COL, colour; WW, weaning weight at 210 days in kg.

2.2. Model of analysis Data were analyzed using a multivariate animal model. The model of analysis for CW, CON, FC and COL was: yijklm = Si + YSj + Hk + SH1 + b⁎AGEm + um + eijklm

ð1Þ

where S was the sex — 2 levels—, YS was the year-season — 35 levels, 3 months per level—, H was the herd — 579 levels—, SH was the slaughterhouse effect — 12 levels— and b was the covariate on age at slaughter (AGE). Moreover, u and e were the random additive genetic and residual effects, respectively. The model of analysis for weaning weight was: yijkl = Si + YSj + Hk + b⁎AR1 + u1 + eijkl

ð2Þ

where AR was the age of recording. For WW, the YS effect included up to 73 levels. A Bayesian multivariate analysis was performed for the five traits with a pedigree of 55,747 animals, 32,511 of whom have recorded data for at least one of the studied traits. The number of equivalent generations (Boichard, 2002) on the pedigree was 4.93. Prior distributions for breeding values and the residuals were assumed multivariate normal with zero mean and variance A ⊗ G and I ⊗ R, where A is the numerator relationship matrix, and G and R are the 5 × 5 matrices of additive genetic and residual variance components. The prior distributions for systematic effects and the (co)variance components were bounded flat uniform distributions. Marginal posterior distribution of each parameter was obtained via integration of multivariate density functions using a Gibbs sampling procedure (Gelfand and Smith, 1990). The implementation of the Gibbs sampler involves sampling from the Gaussian conditional distributions of the systematic effects and breeding values and the inverted Wishart distributions for the genetic and residual (co) variances (Van Tassell and Van-Vleck, 1996). After visual inspection, the analysis was performed with a single chain of one million of iterations, with the first 100,000 being discarded. 2.3. Comparison of selection criteria After (co)variance component estimation, we evaluated the potential consequences of the inclusion of CW on the selection criteria. With this objective, we defined three alternative selection criteria: 1) CW-ST: single trait genetic evaluation on CW 2) WW-ST: single trait genetic evaluation on WW 3) CW-MT: multiple trait genetic evaluation on WW and CW. Initially, we performed single and multiple trait genetic evaluations, given the posterior mean estimate for the variance components. We then selected six subsets of individuals. Firstly, three groups of selected sires, SCW-ST, SWW-ST and SCW-MT, consisting of the top 10% living males born in the last two years, ranked with the three selection criteria (CW-ST, WW-ST and CW-MT). And secondly, three groups of selected dams, DCW-ST, DWW-ST and DCW-MT, consisting of the top 25% living females ranked with the same criteria.

J. Altarriba et al. / Livestock Science 125 (2009) 115–120

Later, we defined new variables related to the expected genetic response of the selection objective (CW) for each subset: NK P i=1

RK ðCWÞ =

uiðCWÞ

NK

NT P

−

j=1

NT

where RK(CW) is the expected selection response in CW for the kth subpopulation, NK is the number of individuals at the kth subpopulation, and NT is the total number of selection candidates. We also defined the standardised response for each criterion as: SRK ðCWÞ =

RK ðCWÞ σ pðCWÞ

where σp(CW) is the phenotypic standard deviation for CW. In addition, expected correlated response (CR) and standardised correlated response (SCR) on CON, FC, COL, and WW were calculated using NK P

CRK ðSÞ =

i=1

uiðSÞ

NK

NT P

−

j=1

ujðSÞ

NT

where CRK(S) is the correlated expected selection response for the kth subpopulation and the sth trait. Finally, we also defined the standardised correlated response for each of the subpopulations as SCRK ðSÞ =

Table 2 Standard deviation among level solutions of each systematic effect and percentage of variance with regard to overall systematic effects (in brackets). Variable a

ujðCWÞ

RK ðSÞ σ pðSÞ

where σp(S) is the phenotypic standard deviation for the sth trait. The posterior predictive distribution for RK(CW), SRK(CW), CRK(S) and SCRK(S) were computed within a global Bayesian analysis with the Gibbs sampler. 3. Results and discussion A summary of the descriptive statistics for the five traits studied is presented in Table 1. These results indicate that the average carcass of the Pirenaica breed under commercial conditions weights around 300 kg at 12 months of age (364.4 ± 40.4 d), with an average conformation score of 3.60, between R (good) and U (very good), a slight fat cover (2.16) and a pale meat colour (2.26). These are the first results for carcass traits obtained under commercial conditions in the Pirenaica Breed (production system, feeding, handling, etc), and they are similar to those previously reported by Piedrafita et al. (2003) and Altarriba et al. (2005) under an experimental environment. In addition, when compared with other published results from the EUROP system of carcass classification in commercial conditions, the Pirenaica breed showed a higher CON and a smaller FC than each of the eight Irish populations studied by Hickey et al. (2007), and the Swedish Simmental population reported by Eriksson et al. (2003). Additionally, their results were comparable to the outcome from the Swedish Charolais and Hereford populations from the same study.

117

CW

CON

FC

Sex 34.1 (68) 0.084 (7) 0.103 (8) Age 14.9 (13) 0.032 (1) 0.029 (1) Trimester-year 4.6 (1) 0.106 (12) 0.094 (6) Herd 17.5 (18) 0.144 (21) 0.156 (18) Slaughterhouse 2.8 (b 1) 0.259 (59) 0.303 (67)

COL

WW

0.035 (4) 0.052 (12) 0.033 (4) 0.100 (41) 0.098 (39)

18.5 (21) 26.2 (41) 8.4 (4) 23.7 (34) –

a CW, cold carcass weight in kg; CON, conformation; FC, fat cover; COL, colour; WW, weaning weight at 210 days in kg.

Furthermore, the COL results were also similar to those reported by Altarriba et al. (2005) in experimental conditions, with paler outcomes than the results obtained by Shojo et al. (2006) for Japanese breeds. Finally, average WW was 264.8 kg, consistent with the results reported by Varona et al. (1997) in the same population (286.2 kg in males and 248.4 in females, at 205th day). In Table 2, the percentages of total variance explained by each systematic effect, in models 1 and 2, are presented. The sex effect was the most important in CW (68%), whereas the slaughterhouse effect captured a bigger percentage in CON (59%), FC (67%) and COL (39%). This fact reveals the complexity of normalization of carcass evaluation, mostly for CON and FC. An intermediate situation was presented for herd effect (18–41%), as well as for the age effect in CW and COL. Finally, for WW, sex (21%), age (26%) and herd (24%) have a similar influence on the total variation explained by the systematic effects. The posterior mean estimates for each level of sex and age reflected the expected physiological relationship among them. Specifically, for the sex effect, males showed bigger estimates than females in CW (71.0 kg), CON (0.17 points), COL (0.08 points) and WW (37.1 kg), with this result being very similar to that of Varona et al. (1997) of 37.8 kg at 205 days. The situation was reversed for FC where females showed a higher fat cover than males (0.21), due to their higher precocity (Warris, 2000). Moreover, for age effect, an almost linear growing relationship was observed for all traits, with average increments of 0.35 and 1.14 kg per day of age in CW and WW, and 0.037, 0.024 and 0.022 points per month for CON, FC and COL, respectively. For season, herd and slaughterhouse effects, there was substantial variation as outlined before, but no clear tendency was detected. Table 3 presents the posterior mean estimates and the Highest Posterior Density at 95% for variance components and heritabilities from the multivariate Bayesian analysis. The posterior estimates for the heritabilities indicate that a relevant additive genetic variance exists for all traits, confirming the results obtained by other studies for CW, CON and FC (Eriksson et al., 2003; Hickey et al., 2007). They found a great heterogeneity in the heritability estimates between breeds, and the results presented here are within their range of estimates. Moreover, the posterior mean estimates of the heritability for CW (0.34) and FC (0.19) were coherent, but lower than the results reviewed by RiosUtrera and Van-Vleck (2004), who reported an average estimate of 0.40 for carcass weight and 0.36 and 0.37 for back fat thickness and marbling score, respectively. The posterior estimate for COL was 0.23, smaller than the results

118

J. Altarriba et al. / Livestock Science 125 (2009) 115–120

Table 3 Estimates of the mean and bounds of high posterior density intervals at 95% (in brackets) of the marginal posterior distribution for the variance components and the heritabilities. Trait a

σ2a

σ2e

h2

CW

436.2 (367.6 504.3) 0.043 (0.034 0.054) 0.030 (0.024 0.037) 0.023 (0.016 0.032) 728.1 (637.8 817.8)

847.3 (797.8 898.2) 0.112 (0.104 0.119) 0.126 (0.120 0.131) 0.076 (0.069 0.083) 1170.0 (1105.7 1286.3)

0.340 (0.291 0.386) 0.276 (0.221 0.340) 0.193 (0.154 0.236) 0.233 (0.162 0.315) 0.383 (0.342 0.424)

CON FC COL WW

a CW, cold carcass weight in kg; CON, conformation; FC, fat cover; COL, colour; WW, weaning weight at 210 days in kg.

obtained by Shojo et al. (2006) in the Hyogo and Tottoti Japanese Breeds. Finally, the posterior mean estimate for WW heritability was also similar with the previous results obtained in the same breed (Varona et al., 1997) and within the range of previous estimates in other beef cattle breeds (Koots et al., 1994a). A summary of the posterior distributions for genetic and residual correlations is presented in Table 4. The posterior HPD at 95% for the genetic correlation did not include zero in the pairs CW-WW, FC-WW, CW-CON, and CON-FC. The genetic correlation between CW and WW was high and positive (0.54), confirming the assumptions of the current selection program (Altarriba et al., 1996), which uses WW as the selection criteria to increase CW. Moreover, WW has a positive genetic correlation with FC (0.23), showing that more precocious individuals have a higher fat deposition, whereas the genetic correlation between CW and FC was not different from zero (− 0.05). Finally, CON presented a positive genetic correlation (0.30) with CW. This result is in strong agreement with the estimates obtained by Van der Werf et al. (1998), Hirooka

Table 4 Estimates of the mean and bounds of high posterior density intervals at 95% (in brackets) of the marginal posterior distribution for the genetic correlations in the upper triangle and for residual correlation in the lower triangle. Trait a CW CW

CON

FC

COL

WW

–

0.335 (0.290 0.379) 0.140 (0.097 0.186) 0.069 (− 0.010 0.151) 0.468 (0.411 0.522)

CON

FC

COL

WW

0.301 (0.167 0.436) –

− 0.048 (− 0.218 0.112) − 0.353 (− 0.520 − 0.187) –

0.224 (− 0.031 0.461) 0.001 (− 0.328 0.275) 0.025 (− 0.188 0.242) –

0.536 (0.426 0.650) − 0.040 (− 0.180 0.110) 0.226 (0.062 0.389) 0.036 (− 0.242 0.306) –

− 0.083 (− 0.124 –0.040) 0.109 (0.031 0.195) 0.125 (0.061 0.193)

− 0.003 (− 0.054 0.048) 0.021 (− 0.037 0.079)

0.084 (− 0.029 0.193)

Estimates with posterior intervals not including zero are in bold. a CW, cold carcass weight in kg; CON, conformation; FC, fat cover; COL, colour; WW, weaning weight at 210 days in kg.

et al. (1998) and Parkkonen et al. (2000). In addition, the null genetic correlation between CON and WW indicates that the genetic relationship between both traits is mainly generated by the post-weaning growth. Conformation score (CON) also presented a high and negative genetic correlation (−0.53) with fat cover (FC), similar to the results reported by Hirooka et al. (1998) and Eriksson et al. (2003), but in strong contrast with the results presented by Van der Werf et al.(1998), Parkkonen et al. (2000) and Hickey et al. (2007). Breed differences in carcass traits have been reported (Marshall, 1994; Piedrafita et al., 2003), and it must be noted that negative correlations have been reported in beef cattle breeds (Hirooka et al., 1998; Eriksson et al., 2003) and null or even positive correlations in dairy cattle breeds (Van der Verf, 1998; Parkkonen et al., 2000 and Hickey et al., 2007). In south European beef cattle breeds, Piedrafita et al. (2003) showed that CON is mainly related with muscular development and, at the phenotypic level, muscular development and fatness are strongly and negatively correlated. Meat colour (COL) did not present any genetic correlation with an HPD at 95% different from zero, but the probability of a positive genetic correlation was 0.96 with CW and 0.93 with WW, confirming the results presented by Shojo et al. (2006) in Japanese Black Cattle. In general, the results of the genetic correlations are in agreement with previous estimates in the literature, as reviewed by Koots et al. (1994b). Additionally, posterior HPD for residual correlations differed from zero in CW-CON, CW-FC, CW-WW, CON-COL, CON-WW and CON-FC. The trend and magnitude of the posterior estimates for the residual correlations are in agreement with results of Piedrafita et al. (2003) at a phenotype level. The expected direct and correlated selection responses with the three selection criteria (CW-ST, WW-ST and CWWT), with one generation of selection, are presented in Table 5 and Fig. 1. For CW, the current objective of selection, the average expected response from sires and dam subpopulations was 7.38 ± 2.42 kg (0.20 ± 0.07 SD) with the WW-ST criteria, whereas it increases up to 8.16 ± 2.38 kg (0.21 ± 0.07 SD) for CW-ST and to 11.48 ± 2.18 kg (0.31 ± 0.06 SD) for CW-

Table 5 Posterior mean (and standard deviation) for the expected response (R) and correlated expected response (CR) for the selected subsets of the populations. Trait a

SCW-ST

SWW-ST

SCW-MT

DCW-ST

DWW-ST

DCW-MT

CW

10.12 (2.73) 0.089 (0.032) − 0.040 (0.029) 0.010 (0.031) 0.59 (3.19)

9.19 (3.29) 0.033 (0.038) 0.004 (0.034) − 0.011 (0.038) 24.91 (3.41)

14.13 (2.74) 0.071 (0.032) − 0.017 (0.029) 0.001 (0.033) 17.55 (3.28)

6.19 (2.05) 0.075 (0.025) − 0.033 (0.023) − 0.002 (0.025) − 1.37 (2.31)

5.56 (2.34) 0.038 (0.028) 0.001 (0.026) − 0.019 (0.029) 15.01 (2.44)

8.82 (2.11) 0.064 (0.025) − 0.017 (0.023) − 0.008 (0.027) 10.05 (2.31)

CON FC COL WW

SCW-MT, SCW-ST and SWW-ST consisting of the top 10% living males born in the last two years, given the three selection criteria (ST and MT, single and multiple trait genetic evaluation). DCW-MT, DCW-ST and DWW-ST, consisting of the top 25% living females with the same criteria. a CW, cold carcass weight in kg; CON, conformation; FC, fat cover; COL, colour; WW, Weaning Weight at 210 days in kg.

J. Altarriba et al. / Livestock Science 125 (2009) 115–120

119

Fig. 1. Direct and correlated genetic response (in phenotypic standard deviations) for three alternative selection criteria (CW-MT, CW-ST and WW-ST)+.

MT. With respect to the current selection criteria (WW-ST), the percentage of increase for expected genetic response in CW was 9% using only the data provided by the SIMOGAN database and 56% with the combined use of the data from the breeders association (WW) and the traceability database (CW). It must be noticed that the last procedure uses WW as criterion of selection, but the breeding values for this trait were obtained from a multiple trait approach, given all the available information. The expected correlated genetic response for CON, FC, COL and WW is presented in Table 5. The HPD at 95% did not include the zero for CON with CW-MT and CW-ST criteria and for WW with CW-MT and the WW-ST criteria. To illustrate these results, the average standardised genetic response from sire and dam subpopulations is presented in Fig. 1. With the current criteria of selection (WW-ST), a relevant genetic response of 0.46 ± 0.06 phenotype standard deviations in WW was expected, as this trait was used directly as the selection criteria. This correlated response decreased to 0.32± 0.06 with CW-MT and disappeared with CW-ST. On the other hand, no genetic response was expected in CON (0.07 ± 0.08 SD) when WW-ST was used, whereas a positive correlated genetic responses appeared when CW-ST and CW-MT were used (0.19 ± 0.07 SD and 0.15 ± 0.07 SD, respectively). Finally, no correlated responses were expected in FC and COL with any of the selection criteria studied. Further research on the economic consequences for carcass and meat quality traits must be performed to determine the weights for a future selection index that includes these traits in the selection criteria, which are now available in abundance from traceability information systems.

Pirenaica beef cattle breed using the available information provided by the Spanish traceability program (SIMOGAN). The use of data from this database attempts to solve one of the main problems in practical animal breeding, the current lack of available information beyond the farm level. Until now, the breeding program for the Pirenaica Breed was focused on increasing carcass weight by developing selection on weaning weight (Altarriba et al., 1996). However, the availability of yearling carcass weights and carcass and meat quality scores, at a very low cost, allows to redefine the breeding objective from a broader spectrum. The results here presented indicate that there is available genetic variability for selection by carcass weight, conformation, fat cover and colour traits; and, furthermore, an alternative scheme for directly improving the weight and carcass traits is compared with the current selection scheme of the Pirenaica breed. In the near future, new developments in information systems will include more traits in traceability databases, eg pH, that are currently in development for the Pirenaica breed. As a consequence, the amount of available information will increase considerably and the current lack of information for data recording will no longer be a significant constraint for animal breeding schemes. In fact, it is expected that data mining techniques will be needed to select the appropriate sources of information for selection objectives and, in this case, rank reduction techniques for genetic evaluation, such as those proposed by Kirkpatrick and Meyer (2004) and Meyer (2007), would be a useful strategy.

Acknowledgements 4. Implications All through the present study we have analyzed the chance of implementing new selection criteria for the

This research was financed by Spain's Ministerio de Educación y Ciencia (AGL2007-66147-C02-01/GAN grant) and carried out in collaboration with the Pirenaica Breeders

120

J. Altarriba et al. / Livestock Science 125 (2009) 115–120

Association (CONASPI) and the Spanish Ministry of Agriculture (manager of SIMOGAN database). References Albertí, P., Ripoll, G., Goyache, F., Lahoz, F., Olleta, J.L., Panea, B., Sañudo, C., 2005. Carcass characterization of seven Spanish beef breeds slaughtered at two commercial weights. Meat Sci. 71, 514–521. Altarriba, J., García Cortés, L.A., Moreno, C., Varona, L., 1996. Situación y perspectivas de la mejora genética de la raza vacuna Pirenaica. ITEA 92A, 107–116. Altarriba, J., Varona, L., Moreno, C., Yagüe, G., Sañudo, C., 2005. Consequences of selection for growth on carcass and meat quality in Pirenaica cattle. Livest. Prod. Sci. 95, 103–114. Altarriba, J., Varona, L., Moreno, C., Yagüe, G., Pastor, F., 2006. Effect of growth selection on morphology in Pirenaica cattle. Anim. Res. 55, 55–63. BOE (Official Bulletin of the Spanish State), 1999. Regulation of the database SIMOGAN. Off. J. 133, 45888–45891. Boichard, D., 2002. Pedig: a Fortran Packaged for pedigree analysis suited for large populations. In: van der Honing, Y. (Ed.), Proceedings of then 7th World Cong. Genet. Appl. to Livest. Prod., Montpellier, 19–23 August 2002. INRA, Castanet-Tolosan, France, CD-Rom, comm. No. 28-13. CEE no. 2930/81, 1981. Community Scale for the Classification of Carcass of Adult Bovine Animals. Official publications of the European communities, Luxemburg. L-2985. Council Directive 97/12/CE, 1997. Council Directive amending and updating Directive 64/432/EEC on health problems affecting intra-Community trade in bovine animals and swine. Off. J. L 109, 1–37. Eriksson, S., Näsholm, A., Johansson, K., Philipsson, J., 2003. Genetic analyses of field-recorded growth and carcass traits for Swedish beef cattle. Livest. Prod. Sci. 84, 53–62. Gelfand, A., Smith, A.F.M., 1990. Sampling-based approaches to calculating marginal densities. J. Am. Stat. Assoc. 89, 398–409. Hickey, J.M., Keane, M.G., Kenny, D.A., Cromie, A.R., Veerkamp, R.F., 2007. Genetic parameter for the EUROP carcass traits within different groups of cattle in Ireland. J. Anim. Sci. 85, 314–321. Hirooka, H., Groen, A.F., Van der Werf, J.H.J., 1998. Estimation of additive and non-additive genetic parameters for carcass traits on bulls in dairy, dual purpose and beef cattle breeds. Livest. Prod. Sci. 54, 99–105. Kirkpatrick, M., Meyer, K., 2004. Direct estimation of genetic principal components: simplified analysis of complex phenotypes. Genetics 168, 2295–2306.

Koots, K.R., Gibson, J.P., Smith, C., Wilton, J.W., 1994a. Analyses of published genetic parameters estimates for beef production traits 1. Heritability. Anim. Breed. Abstr. 62, 309. Koots, K.R., Gibson, J.P., Smith, C., Wilton, J.W., 1994b. Analyses of published genetic parameter estimates for beef production traits. 2. Phenotypic and genetic correlations. Anim. Breed. Abstr. 62, 825–853. Marshall, D.M., 1994. Breed differences and genetic parameters for body composition traits in beef cattle. J. Anim. Sci. 72, 2745–2755. Meyer, K., 2007. Multivariate analysis of carcass traits for Angus cattle fitting reduced rank and factor analytic models. J. Anim. Breed. Genet. 124, 50–64. Parkkonen, P., Liinamo, A.-E., Ojala, M., 2000. Estimates of genetic parameters for carcass traits in Finnish Ayrshire and Holstein–Friesian. Livest. Prod. Sci. 64, 203–213. Piedrafita, J., Quintanilla, R., Martín, M., Sañudo, C., Olleta, J.L., Campo, M.M., Panea, B., Renand, G., Turin, F., Jabert, S., Osoro, K., Oliván, C., Noval, G., García, M.J., Garcıía, D., Cruz-Sagredo, R., Oliver, M.A., Gil, M., Serra, X., Guerrero, L., Espejo, M., García, S., López, M., Izquierdo, M., 2003. Carcass quality of 10 beef cattle breeds of the Southwest of Europe in their typical production systems. Livest. Prod. Sci. 82, 1–13. Renand, G., Larzul, C., Le Bihan-Duval, E., Le Roy, P., 2003. Genetic improvement of meat quality in the different livestock species: present situation and prospects. Prod. Anim. 16, 159–173. Rios-Utrera, A., Van-Vleck, L.D., 2004. Heritability estimates for carcass traits of cattle: a review. Genet. Mol. Res. 3, 380–394. Sanchez, A., Ambrona, J., Sanchez, L., 2002. Razas ganaderas españolas bovinas. Ed. MAPA-FEAGAS, Madrid. Shojo, M., Okanishi, T., Anada, K., Oyama, K., Mukai, F., 2006. Genetic analysis of calf market weight and carcass traits in Japanese Black cattle. J. Anim. Sci. 84, 2617–2622 2006. Van der Werf, J.H.J., Van der Waaij, L.H., Groen, A.F., de Jong, G., 1998. An index for beef and veal characteristics in dairy cattle based on carcass traits. Livest. Prod. Sci. 54, 11–20. Van Tassell, C.P., Van Vleck, L.D., 1996. Multiple-trait Gibbs sampler for animal models: flexible programs for Bayesian and likelihood-based (co) variance component inference. J. Anim. Sci. 74, 2586–2597. Varona, L., Moreno, C., García Cortés, L.A., Altarriba, J., 1997. Multiple trait genetic analysis of underlying biological variables of production functions. Livest. Prod. Sci. 47, 201–209. Warris, P.D., 2000. Meat science. An introductory text. CABI Publishing, London, UK.